As opposed to European country funds, where the commercial target market is the home market, the upcoming European sector funds could, in theory, be launched on a European scale. If the fund fulfils certain diversification criteria, it receives a European password with which it can commercialise its fund in all European countries without additional controls. But market capitalisation-based sector indexes fail to comply. So do correlation-based indexes hold the solution?
The popularity of sector-based investment over country-based investment in Europe is increasing every day. We can illustrate this by examining the situation in Belgium. Graph 1 compares the total number of new sector funds with the number of new country funds (within Europe) launched during the last five years. The different pattern is clear. This trend can be extrapolated to the rest of Europe. Another indication of the rising interest in European sectors was EuroNext’s recent declaration that it would launch exchange traded funds (ETF’s), not only on general European indexes but also on European sector indexes.
The switch from country-based investing to a more sector-based approach, brings with it some nice opportunities for the fund industry. Financial institutions with a presence in several European countries can now, say, establish a technology-sector fund in Germany but commercialise it on a
European scale. Trying the same thing a couple of years ago with a passively-
managed country fund based on, for
example, the German DAX would never have worked. The same goes for ETFs. It would be ideal to launch an ETF on EuroNext and commercialise it all over Europe. But however nice the economies of scale might seem in theory, reality still
presents some obstacles.
The major problem fund managers with international ambitions face is the European passport. A fund created in one member of the European Union can be sold in the other European countries without additional controls if it complies with the rules set out in European Directive 85/611/EEC of 20 December 1985. According to this directive, an investment fund may invest no more than 5% of its assets in transferable securities issued by the same body. This limit is raised to an absolute maximum of 10%. However, the total value of the shares held by a fund in which it invests more than 5% of its assets in each must not exceed 40% of the value of its assets. Complying with the 5/10/40-directive is a guarantee for a highly diversified portfolio of shares.
For an investment fund linked to a pan-European index, the directive rarely produces any problems. In the case of the Euro Stoxx 50, for example, only three stocks have a weighting of more than 5%,
namely Nokia, Royal Dutch Petroleum and TotalFina Elf. But creating index-linked funds or ETF’s in European sectors that
fulfil the 5/10/40-directive criteria and therefore can be sold in every European country is a very different matter. The
problem stems from the greater difference in market capitalisation between the
constituents of a pan-European sector index and a general pan-European index. Given that the most popular indices are
currently capitalisation based – corrected for free float or not – big differences in
market capitalisation will always lead to a poorly diversified and an unevenly weighted index.
The table illustrates the problem. Starting from the biggest and most liquid stocks of six European sectors, we calculated the theoretical weighting of the constituents in the different sector indexes based on their actual market capitalisation. Having created market-cap weighted sector indexes, we than checked whether they complied with the 5/10/40-rule. The result is troubling. All of the indexes break the 40%-rule and all but one breach the absolute maximum weighting limit of 10% for one share. The last one is the financials index with no less than 69 constituent shares. Is there an easy way around this problem? Introducing a cap of 10% solves the difficulty of the 10% rule, but does not help with the 40% rule. What are the alternatives?
Jan Geerts, head of index fund management at KBC Asset Management, does not see any alternative arising in the near future. “A solution would be to track the index via futures, but this option was rejected by the controlling banking organism,” he says. “I don’t see any other way around this problem, so what we did at KBC with sector funds – and what is common practice in Europe – is stop trying to comply with the 5/10/40-rule.
“The only thing left at that moment is to comply with the specific rules in force in the different countries where you want to launch your passive managed sector fund,” Geerts adds.
Does this means that the European directive has not been thought through? From the point of view of ‘protecting the small investors’ the directive is a good thing. For some sectors, however, complying with the directive is impossible. European sector indexes for the automobile or oil industry have less than 20 potential constituents if some minimum liquidity criteria are taken into account. For most of the other sectors, however, an alternative is available. As stated before, the big problem with the most known European sector indexes is poor distribution due to capitalisation-based weighting.
Using this methodology with European sector indexes would mean many stocks would have to be included in the indexes before they comply with the directive. As shown in the table, even with 60 stocks the financials index does not fulfill the requirements. The only consistent solution to the problem is to employ a different weighting methodology.
Correlation-based weighting might be one possible answer. Here the starting point is not that the biggest stock gets the biggest weighting, but that the stock that best tracks the sector trend will receive the highest weighting. In other words, the weightings of the constituent stocks in the index are a function of their correlation with the sector trend. Why would this give a better distribution of the weightings? A clear sector trend will only emerge if a good many stocks tend to move for a long period in the same direction, or in other words are highly correlated. At that moment the correlation from one stock to the other won’t differ greatly, what will be reflected is the weighting of the index.
The theory sounds nice, but what about in practice? Graph 2 compares the ratio ‘weighting lightest stock/weighting heaviest stock in index’ of market cap based sector indexes versus correlation-based indexes. In the first category, the weighting of the smallest stock was on average 100 times smaller than that of the biggest one. In the case of correlation-based weighing, the largest weighted stock never exceeds twice the weighting of the smallest stock. The best example of applicable methodology is the financials index or IN.fin. This index perfectly fulfils the 5/10/40-requirement with only 28 stocks. None of the stocks exceed a weighting of even 5%. Geerts argues that this is a workable solution. “You have to be convinced of the entire concept of correlation-based weighting,” he says.

Koen De Leus is chief analyst at European Benchmarks in Antwerp